Talk CoTurn running on separate VM vs Directly On the Nextcloud VM

ℹ️ Support 💬 Talk (spreed)
,
1

Hi,
Curious about something.
Talk CoTurn running on separate VM vs Directly On the Nextcloud VM

I’m confused about why I setup a separat VM for Coturn server.
The Nextcloud VM and the Coturn Vm are on the same server at the moment.
I’m not sure it matters or why I did it this way. Perhaps I was afraid of messing up my nextcloud install at the time. I don’t really remember.

Is there any real reason to have my Coturn server running on separate VM ?
And what is meant by a high end backend for TALK STun / Coturn etc. ? What does an example of this high end look like so I can understand an example configuration for this topic.

Thanks

2

Hey, it is referring to this: https://github.com/strukturag/nextcloud-spreed-signaling There are quite some topics and posts about this here, just take a look.

There’s no technical reason per se, if that is your question, but there might be some reasons in general. I’d say it’s mostly about separation:

  • Different resource allocation
  • Separation of concerns
  • Different server (maybe a different IP if you want to use the same ports)
  • Different firewall rules
  • etc
3

Thanks
I installed Coturn on VM based on some assumptions that I made during initial setup and also how to get Talk Working etc.

My assumption is that Talk doesn’t work out of the box with nextcloud and was advised early that a Coturn server is needed which I also installed and Talk was working as expected at that time.

In the NextCloud:Admin section for (Talk) There is the section for WebServer Setups etc.

Coturn section has my settings and listed as (turn:only).
I don’t know the difference between “turn only and turn and turns”.

There there are more then one signal server option listed in the admint/talk for high end backup such as:
High-performance backend
Hosted high-performance backend
Recording backend

I’m not exactly sure which of these I need or do not need.
Which is relevent or is all of them relevent.
I can’t seem to find out what are the differences.

4
  • High-performance backend: is the name given to the backend. the “high performance” part depends on usage and resources
  • Hosted high-performance backend: the backend is hosted by a third party, be it behind a paywall or “free” requiring an email or personal data
  • Recording backend: an extra backend to enable recording meetings, requires disk space for keeping the recordings

:rofl: yeah, neither do I

this is how its documented in my personal notes… and I’m no boffin, just trying to understand…

if you require external video conferencing and voice chats between 5 or more WebRTC devices/participants, you’d need a TURN instance to organise the transmission of the data stream between WebRTC devices. (reduces video choppiness)

the “s” usually implies SSL or TLS (encrypted) communication. some third party HPB-providers offer encryption, but since usually Nextcloud is encrypted anyway, Talk will be encrypted too. So turns is seldom needed.

this is commonly known as “signalling” which basically handles the connection between the devices in a NATed environment making sure that the IP’s of each device is known and can be connected.

depending on your use case

  • you’ll need TURN for 5 or more external WebRTC devices or participants
  • you’ll need Talk:HPB (high performance backend) also known as “signalling” to connect NATed external IP’s
  • you’ll need recording if you need to make video recordings of the conferences

See my personal wiki: wiki-md/NEXTCLOUD.snap--Talk_HPB-with-docker.md at scubamuc-wiki · scubamuc/wiki-md · GitHub

1 Like